Skip to main content
POST
Submit an image for quality evaluation

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Headers

X-Organization-Id
string | null

Body

application/json
generated_image_url
string
required

The image to evaluate: a public https:// URL, a runflow://assets/{uuid} reference from POST /v1/asset-uploads, or a data:image/... URI (materialized into a hosted asset).

Maximum string length: 12000000
task_type
string
required

What kind of generation produced the image, e.g. headshot or image_generation.

Required string length: 1 - 100
task_description
string | null

Longer description of the generation task, used to plan the evaluation.

Maximum string length: 4000
generation_prompt
string | null

The prompt that produced the image — strongly recommended for prompt-adherence judging.

Maximum string length: 10000
evaluation_instructions
string | null

Free-text rubric: what the evaluator should prioritize or ignore.

Maximum string length: 4000
input_attributes
Input Attributes · object | null

Structured generation attributes (e.g. {"style": "corporate"}) shown to the evaluator.

reference_images
EvaluationReferenceImage · object[] | null

Reference images for identity/garment comparison (max 4).

Maximum array length: 4
run_id
string<uuid> | null

Optional: associate this evaluation with one of your Runflow runs. The submitted image does not need to be the run's output (e.g. a post-processed export). Platform auto-evaluation of the run's own output is unaffected.

client_ref
string | null

Your correlation label, echoed in responses and callbacks. Not an idempotency key.

Maximum string length: 255
job_class
string
default:standard

Evaluation job class — see GET /v1/evaluations/job-classes for active classes and per-class prices. Currently only standard.

Maximum string length: 50
callback_url
string | null

HTTPS endpoint to receive a signed evaluation.completed / evaluation.failed callback when the evaluation terminates.

Maximum string length: 2048

Response

Successful Response

id
string<uuid> | null
org_id
string<uuid> | null
run_id
string<uuid> | null
target_type_code
string | null
model_id
string<uuid> | null
flow_id
string<uuid> | null
comfyui_workflow_id
string<uuid> | null
status_code
enum<string> | null
Available options:
pending,
running,
completed,
failed
task_type
string | null
task_description
string | null
generation_prompt
string | null
generated_image_url
string | null
input_attributes
Input Attributes · object | null
evaluation_instructions
string | null
reference_images
Reference Images · object[] | null
overall_passed
boolean | null
weighted_pass_rate
number | null
override_reason
string | null
top_issues
string[] | null
top_strengths
string[] | null
check_summary
Check Summary · object | null
analysis_modules_executed
integer | null
primary_action_code
string | null
parent_evaluation_id
string<uuid> | null
attempt_number
integer | null
superseded_at
string<date-time> | null
eval_duration_ms
integer | null
feedback_is_positive
boolean | null
feedback_reason
string | null
created_at
string<date-time> | null
updated_at
string<date-time> | null
source_code
enum<string> | null
Available options:
platform,
api
client_ref
string | null
submitted_at
string<date-time> | null
completed_at
string<date-time> | null
callback_url
string | null
cost
string | null
Pattern: ^(?!^[-+.]*$)[+-]?0*\d*\.?\d*$
failure_code
enum<string> | null
Available options:
invalid_media,
dispatch_failed,
processing_failed,
timed_out
job_class_code
enum<string> | null
Available options:
standard,
quick
judges
FullValidator · object[] | null
gate_failures
FullValidator · object[] | null
action
PublicValidator · object | null
run
FullValidator · object | null

Response shape.

model
FullValidator · object | null

Response shape.

parent_evaluation
unknown